The Problem

When we first hooked up the power for the circuit board, we were getting feedback from the power LEDs, and all of the power rails appeared to be normal. However, when we put a 10KHz sine wave of 1V amplitude into the BNC input, all we were getting out on the other side was noise. This prompted us to revisit the design of the analog section of the board. In the original design, the input signal to the BNC is attenuated and DC biased in order to go into the filter IC. However, we realized that the attenuator circuit is improperly designed as it did not take into account the loading effects of the much lower impedance filter components.  This caused our attenuation to be much higher than we expected and therefore, we needed to modify the circuit.

The Fix

After attempting to fix the problem by swapping resistor values in simulation, we realized that we needed a more drastic design modification.  Alan had a good idea for a fix that would slightly degrade performance but allow the circuit to work as we had originally intended. In short, we removed most of the components around the filter and used a single op amp inside of the filter IC as a unity gain, high-impedance buffer. This fix proved to be very effective. Using the filter as a single op amp, we were able to maintain our 20:1 attenuation and a DC bias of 1.65V, which were both intended in the original design. After the buffer output, we added a simple RC filter with a corner at 4.8MHz to reduce high-frequency noise. The corrected circuit and our jumper wire fix are shown below.